The number of datasets and problems involving both location and time is growing rapidly with the increasing availability and importance of large spatio-temporal datasets such as GPS trajectories, climate records, social networks, sales transactions, etc. Advancement of machine learning and knowledge discovery methods for such datasets is critical for the development of smart cities, addressing climate and ecological challenges of this century, controlling disease outbreaks, etc. In particular, good understanding of spatio-temporal datasets is vital for degradation studies in material science, which lies at the core of sustainable economies.
The Ph.D. student will be based at the Computer Science department (a top in research department within UK, based on REF 2014) and will work in close collaboration with the R&D division of a large international company, which will provide material geo-weathering datasets from multiple locations around the globe for the last ten years. The project aims to apply and improve state of the art techniques for spatio-temporal analysis, regression and clustering. The student will leverage both classical and kernel-based learning methods as well as neural networks and deep learning depending on the problem at hand. As the developed methods will be eventually used in the industrial context, the project also incorporates uncertainty estimations and explainable AI contributions.
This project form a part of the Doctoral Training Centre for Next-Generation Materials Chemistry (https://www.liverpool.ac.uk/study/postgraduate-research/studentships/next-generation-materials-chemistry/). This new Centre that aims to deliver a new cross-disciplinary approach to materials chemistry research. The Centre will train PhD graduates at the interface of physical science, AI, data science, and robotics to create the leaders in data-enabled science that UK industry and academia requires to deliver R&D 4.0. We seek applicants with a strong undergraduate background in mathematics, computer science, engineering, or physics for this post. The 42 month Ph.D project will tackle multidisciplinary problems which were co-defined by the industrial partner working with University of Liverpool academics in the physical sciences and computer science. Core training in data science will be provided together with leadership and entrepreneurship development.
Students in the Doctoral Training Centre for Next-Generation Materials Chemistry (https://www.liverpool.ac.uk/study/postgraduate-research/studentships/next-generation-materials-chemistry/) will be located in the newly opened Materials Innovation Factory (MIF - https://www.liverpool.ac.uk/materials-innovation-factory/), which collocates academic and industrial researchers over 4 floors, with state-of-the-art automated research capabilities, including the £3M Formulation Engine. They will benefit from the cross-disciplinary training environment of the MIF, which contains staff from Physics and Computer Science as well as Chemistry, and the well-established community around the Leverhulme Research Centre in Functional Materials Design (https://www.liverpool.ac.uk/leverhulme-research-centre/), which is typified by a vibrant functioning engagement between physical science and computer science. Industrial partners include Unilever, Johnson Matthey and NSG Pilkington. Supervision is provided from both Chemistry and Computer Science, with the exact make-up of the supervisory team tailored to the students undergraduate background.
The projects address the application of machine learning to inorganic materials chemistry. The inorganic materials chemistry group, led by Prof Rosseinsky at the University of Liverpool (https://www.liverpool.ac.uk/chemistry/research/rosseinsky-group/about/), focusses its research on the discovery of new solid inorganic compounds. Recently, the use of computational materials chemistry has accelerated this materials discovery process, leading to the synthesis of a range of novel metal oxides with a variety of functional properties. These successes have shown that the process of computer aided materials discovery relies on a close working relationship between computational and experimental researchers within the group, which is recognized in the EPSRC Programme Grant in Integration of Computation and Experiment for Accelerated Materials Discovery, and the decision to bring together theoretical and experimental researchers within the Materials Innovation Factory and the Leverhulme Centre for Functional Materials Design at the University of Liverpool. The successful candidate will participate in this relationship, using the development of computational models to guide experimental work and thus accelerate the discovery of new materials.
Applications are welcomed from students with a 2:1 or higher master’s degree or equivalent in Mathematics, Computer Science, Physics, or Materials Science, particularly those with some of the skills directly relevant to the project outlined above. Successful candidates will have strong math and programming skills.
Informal enquiries can be addressed to Vladimir Gusev (firstname.lastname@example.org) or Yannis Goulermas (email@example.com).
Tel. No. for Enquiries: +44 (0)151 794 4520
Please apply by completing the online postgraduate research application form here: https://www.liverpool.ac.uk/study/postgraduate-research/how-to-apply/
Please ensure you quote the following reference on your application: University of Liverpool Doctoral Training Centre in Next-Generation Materials Chemistry CDT03
Applications should be made as soon as possible.
Open to EU/UK applicants
The award will pay full tuition fees and a maintenance grant for 3.5 years. The maintenance grant will is £ 15,285+ for 2020-2021 with additional travel and equipment support. The award will pay full home/EU tuition fees and a maintenance grant for 3.5 years. Non-EU applicants may have to contribute to the higher non-EU overseas fee.
Deep Learning for Spatio-Temporal Data Mining: A Survey. https://arxiv.org/abs/1906.04928
Spatio-Temporal Data Mining: A Survey of Problems and Methods https://doi.org/10.1145/3161602
Deep Learning for Time-Series Analysis https://arxiv.org/pdf/1701.01887.pdf
Transfer learning for time series classification https://arxiv.org/pdf/1811.01533.pdf